Human Ecology, Vol. 28, No. 4, 2000
The Landscape Ecology of Pastoral Herding:
Spatial Analysis of Land Use and Livestock
Production in East Africa
Peter B. Coppolillo1,2
Understanding landscape-scale patterns of herding is critical in identifying
and assessing the impacts of pastoral grazing. Here, a general model of
herding is developed based on the Sukuma agropastoral system in the Rukwa
Valley, Tanzania. Using this conceptual framework, the factors affecting the
maximum distances herds travel from home and the distribution of grazing
around pastoral settlements are examined. The distribution of dry season
water structured the landscape-scale distribution of grazing throughout the
year, not just during the dry season. Water availability strongly affected the
distances herds ranged from home in the dry season and the distribution of
grazing around pastoral settlements throughout the year. Associations between cattle productivity and herding practices were also examined. The
effects of traveling further from home, keeping cattle in large herds, and using/
living in areas of high settlement densities were examined on the following
measures of productivity: intake rates, foraging behavior, milk yields, and
body conditions. Cattle from larger herds were observed to walk more while
actively foraging and engage in more walking bouts (taking ten steps without
taking a bite). The increased walking of large herds may explain why they
range farther from home and highlight the importance and ubiquity of herd
splitting among pastoralists. However, herd size effects were not apparent
in intake rates or milk yields. Milk yields were negatively affected by traveling
farther from home. These data demonstrate substantial variability within
herding populations and show interesting similarities with herding systems
in substantially more arid areas.
KEY WORDS: pastoralism; GIS; Sukuma; Tanzania.
1
Department of Anthropology, Graduate Group in Ecology, University of California, 1 Shields
Ave., Davis, California 95616-8522. Fax: 530 752-8885, [email protected]
2
Current address: Wildlife Conservation Society, International Programs, 2300 Southern Blvd.,
Bronx, New York 10460-1090. Fax: 718 220-7114, [email protected]
527
0300-7839/00/1200-0527$18.00/0  2000 Plenum Publishing Corporation
528
Coppolillo
INTRODUCTION AND BACKGROUND
Identifying and characterizing the ecological impacts of pastoral production are contentious topics in both the anthropological and biological
literature (Brown, 1971; du Toit and Cumming, 1999; Ellis and Swift, 1988;
Enghoff, 1990; Homewood, 1987, 1995; Homewood and Rodgers, 1984,
1991; Lamprey, 1983; Lane, 1996; Lindsay, 1989; Little, 1996; McCabe,
1990b; Nyerges, 1980; Prins, 1992; Sinclair and Fryxell, 1985; Ward et al.,
1998). A key element of these debates is the spatial dimension of impacts
and pastoral herding, i.e., where herding takes place and what factors affect
this distribution. Anthropologists have long recognized the importance of
mobility and the spatial aspects of pastoral land use (reviewed in DysonHudson and Dyson-Hudson, 1980, and McCabe, 1994). Like other grazing
systems, pastoral herding and the decisions that affect it occur across a
hierarchy of spatial and temporal scales (see Bailey et al., 1996, Coughenour,
1991, and Senft et al., 1987, for general reviews of grazing systems). These
range from continental-scale expansion and migration (Collett, 1989; Johnson, 1989; Marshall, 1990; Waller, 1985), to nomadic movements (DysonHudson & Dyson-Hudson, 1980; McCabe, 1985, 1994), more regular seasonal transhumance (Evans-Pritchard, 1940; Western, 1975; Nyerges, 1982;
Århem, 1985; Homewood and Rodgers, 1991), daily herding (Coppock et
al., 1986a, 1986b; Homewood and Rodgers, 1991; Nyerges, 1982) down to
sub-daily, landscape- or patch-scale movements (de Boer and Prins, 1989).
At each of these levels of organization, a variety of factors influence
the distribution of pastoralists and their livestock. Understanding the factors
affecting the distribution of grazing (and their relative importance) at any
of these scales may help to recognize heterogeneity in resource use (e.g.,
Coppolillo, in press), point to areas where resource conflicts are likely or
particularly acute (e.g., Fox et al., 1996), or provide a deeper understanding
of the long-term factors shaping current landscape structure and ecosystem
function (e.g., Turner, 1998a, 1998b).
Here I present a conceptual model of daily herding and describe a
method for using a Geographic Information System (GIS) to evaluate the
factors affecting daily herding patterns and their resultant effects on livestock productivity. Using this conceptual framework, I examine the factors
affecting households’ daily patterns of pastoral grazing and how variation
in these patterns affects herd productivity among Sukuma agropastoralists
living in western Tanzania. In the first analysis I examine the effects of
labor availability, cultivation practices, herd size, water availability, and
settlement density on the distances herds traveled from home and the
distribution of grazing around herding households. I then examine different
land use practices and their effects on alternative measures of herd produc-
Landscape Ecology of Herding
529
tivity. Specifically, I analyze the effects of distances traveled from home,
settlement density, and herd size on foraging behavior (intake and stepping
rates), milk yields, and animals’ body condition.
The perspective of this study differs from previous research in two
ways. First, rather than characterizing the overall Sukuma herding system,
I focus on variability within the system. Many (but not all) of the studies
mentioned above analyze and describe herding at the population level.
While population-level approaches have undoubtedly proven fruitful, providing the foundation for our current understanding of pastoral systems,
one limitation is that they do not examine variability among individual
herding units. Examination of individual-level variability in land use practices can provide a deeper understanding of the factors affecting herding
decisions (Dyson-Hudson and Dyson-Hudson, 1969; Borgerhoff Mulder &
Sellen, 1994; McCabe, 1994; Sieff, 1997) as it has for other aspects of pastoral
systems (e.g., Borgerhoff Mulder, 1998; Mace, 1993a, 1993b; Roth, 1996;
Sellen, 1999). Second, I examine daily herding through the lens of landscape
ecology. By focusing on the relationships between landscape pattern and
ecological processes (Urban et al., 1987; Forman, 1995), landscape ecology
offers a useful framework for examining the factors affecting daily herding
and how they may translate into overall landscape patterns (Coppolillo, in
press; McIntyre and Hobbs, 1999). This perspective and its focus on the
relationships between herding and specific landscape features (e.g., water
and other pastoral settlements) builds on previous studies of herding
(Homewood and Rodgers, 1991; Nyerges, 1982; Western, 1975), but is also
a departure from others, which focus on economic (e.g., Sandford, 1982;
Schneider, 1957), social (Conant, 1982; McCabe, 1990b), and political (Johnson, 1989) determinants of land use. The approach presented here serves
two purposes. First, it provides a quantitative method for analyzing and
predicting the spatial distribution of pastoral grazing. This is important as
it provides a quantitative foundation for ecological monitoring, a critical
element of natural resource management and conservation projects aimed
at engaging local people (Kremen et al., 1994). Second, it evaluates the
importance of factors affecting the distribution of grazing intensity. This
may be important in understanding the household and landscape factors
underlying observed distributions of grazing intensity.
The Rukwa Valley and the Sukuma Agropastoral System
This study was undertaken in the Rukwa Valley, which is the southern
portion of the Great Rift Valley as it passes through western Tanzania
530
Coppolillo
(Fig. 1). The topography is mostly flat with some rolling hills reaching
elevations around 960 m. Annual precipitation in the Rukwa Valley is
600–900 mm and is generally concentrated in a single rainy season lasting
from early December until early April. Vegetation is a mixture of seasonally
flooded grasslands dominated by Echinochloa pyramidalis, Themeda triandra, and Sporobolus pyramidalis, and Miombo woodland, dominated by
Brachystegia, Julbernardia and Acacia spp. Soils are a mixture of welldrained sandy soils and clayey poorly-drained ‘‘black cotton’’ soils. The
Rukwa Valley is also an area of high conservation value. Home to Katavi
National Park, Rukwa Game Reserve (KNP and RGR, respectively, Fig.
1), and four contiguous Game Controlled Areas, the Valley supports dense
aggregations of large herbivores including buffalo, hippopotamus, elephant,
giraffe, and a full suite of plains game and carnivores.
During the last 20 years the Rukwa Valley has experienced Tanzania’s
highest growth rate outside of the capital city (5.7% per annum in Mpanda
District). Fueled largely by immigration, human population growth has
been accompanied by fivefold increases in cattle numbers and changing
Fig. 1. The location of the study area, Katavi National Park (KNP), Rukwa Game Reserve
(RGR), and the 24 households included in the study.
Landscape Ecology of Herding
531
land uses (Tanzania, 1988, 1989, 1992, 1994). This study was initiated to
understand the current land use system and the constraints that herding
households face in order to inform livestock and natural resource management in the Rukwa Valley and elsewhere.
Research took place among Sukuma agropastoralists, a Bantu-speaking ethnic group who have settled in the Rukwa Valley during the last 25
years. Originally from Shinyanga, Mwanza, and Tabora Regions (to the
north of Rukwa Region), the Sukuma are Tanzania’s largest ethnic group
and are now numerous in every region (Galaty, 1988). The human and
livestock population growth in Rukwa Region and the Rukwa Valley is
largely due to Sukuma immigrants; in fact, in many villages Sukuma now
outnumber the original inhabitants, the Pimbwe and Rungwe (M. Ndegewilaya, pers. comm.). Shinyanga, Mwanza, and Tabora are widely considered
severely degraded (Brandström et al., 1979; Brandström, 1985; Maganga,
1987) and many immigrants cite environmental degradation and a lack
of grazing land as factors driving them from their traditional homeland
(Coppolillo and Borgerhoff Mulder, unpublished data, and see Charnley,
1994, 1997).
The Sukuma of the Rukwa Valley are strongly agropastoral, keeping
mostly cattle with a few goats and sheep. All households grow maize using
ox-drawn plows; poor households without steers or a plow enlist the help
of wealthier relatives or pay a neighbor to prepare their fields. Some households also cultivate millet (for subsistence), and rice (for subsistence and
cash). Cassava, finger millet, and a variety of small vegetable crops are also
cultivated by hand. During the wet season, cattle are herded in fallow fields
and undisturbed areas surrounding Sukuma settlements; harvested stubble
fields are also used during the dry season. Livestock in the Rukwa Valley
are only herded during the day, and all animals are enclosed at night. This
is because hyenas are common and lions occasionally enter the study area
from Katavi National Park, so animals left to graze at night would almost
certainly be preyed upon.
As a result of ‘‘villagization’’ in the late 1970s, settlements in the
Rukwa Valley are nucleated around a central village farm (though these
are no longer communally run, see Kikula, 1997). For the most part, the
central villages are occupied by Pimbwe and Rungwe, the original inhabitants of the Rukwa Valley, and Sukuma immigrants live in more dispersed
satellite settlements surrounding villages. This is in part because of their
recent arrival, but also because livestock conflicts are more common in
densely settled areas, so most herds are kept outside of village centers. By
living on the margins of existing villages, Sukuma are also able to cultivate
larger areas and keep more land in fallow (for soil recovery and grazing),
while still having access to the village for goods and services and social
532
Coppolillo
reasons. Within the satellite areas, the location of Sukuma settlements is
primarily driven by the availability of arable land, forcing some households
to settle as far as four kilometers from water.
Land tenure is relatively flexible in the Rukwa Valley. Land within
village boundaries is titled on 33 year leases, but as mentioned above, most
Sukuma live outside these areas. Here, prospective settlers must be granted
permission to settle, a decision based largely on whether surrounding neighbors consider an area to have sufficient land for cultivation; if there is too
little, the expectation is that land conflicts will be more likely to arise, so
permission is effectively denied by discouraging immigrants from settling.
Sukuma report that in theory, cattle can be herded anywhere not currently
cultivated or recently harvested. In practice however, it is rare for anyone
from more than a few kilometers away to use areas near others’ settlements.
Most households keep all of their cattle at the same settlement where they
cultivate, but a few of the wealthiest herd owners (with ⬎250–300 cattle)
keep cattle at distant settlements under the supervision of older sons. Some
movements of animals do occur between home and distant settlements,
but these generally involve fewer than 10% of animals and are not strongly
seasonal, so they cannot be accurately characterized as transhumance.
Water is managed in a more structured manner than grazing land.
Recall that the topography of the Rukwa Valley is relatively flat; thus during
the wet season (December–April) water is available virtually everywhere in
shallow pools and flooded grasslands, so herding is not constrained by water
availability. During the dry season herders dig shallow ‘‘scratch’’ wells to
reach groundwater in dry river beds, or deeper wells where water must be
hauled up using a bucket and a rope. Access to both kinds of wells is
restricted to the herd owner who dug the well. Herds belonging to close
kin (fathers’ and sons’ herds or brothers’ herds) sometimes share wells,
but this is not always the case. There is also one natural spring in the study
area which is diverted to a pipe (for people) and a trough for livestock.
All herds have access to the spring provided cattle are kept away from the
area around the pipe.
A Conceptual Model of Daily Grazing
In this section I present a conceptual model of grazing based on the
Sukuma agropastoral system described above. The model focuses on daily
patterns of pastoral grazing. I use the term ‘‘grazing’’ broadly, to include
herbivory and its associated ecological processes like mechanical disturbances from trampling and nutrient transfers from dung and urine deposition. Two primary characteristics of daily grazing patterns are the herding
Landscape Ecology of Herding
533
radius and the relative distribution of grazing within it. These two characteristics describe the pattern of grazing around a single pastoral settlement
and, by extension, the landscape-scale distribution of grazing in areas with
multiple settlements.
Herding radius is defined as the maximum distance a herd travels from
home during any single day. Although strongly correlated, herding radius
is not simply half the total distance traveled; if a herd takes a circuitous
route but never strays farther than 1 km from home, the herding radius is
1 km regardless of how far it traveled in total. While the overall distance
that a herd travels (both horizontally and vertically) is undeniably important
in terms of livestock energetics and productivity (Coppock et al., 1986a;
Coppock et al., 1986b; Homewood and Rodgers, 1991; but see Western
and Finch, 1986), this study focuses on herding radius because it circumscribes the area used by each herd. This, in turn, affects the degree of
overlap in grazing by herds from different pastoral settlements and drives
landscape-scale heterogeneity in grazing intensity.
A variety of household-level factors may influence herding radius.
First, the need to water animals may draw herds farther from home than
would be necessary based solely on grazing resources. Alternatively, being
close to water may eliminate travel time to and from watering and allow
a herd to travel farther from home. Thus, positive and negative relationships
between herding radius and households’ distance to water are possible.
Second, herds from households surrounded by larger numbers of other
pastoral settlements may be forced to travel farther to find sufficient grazing
resources, so settlement densities may also affect herding radius. Third,
households’ labor pools and cultivation practices may affect herding radius
by constraining the amount of time allocated to herding and therefore, the
distance herds are able to travel from home. The effects of labor may
depend not just on the number of people in the household, but also on the
demand for their labor. In this case an interaction between labor and
cultivation would affect the amount of time allocated to herding. Looking
only at cultivation, herds from households with larger areas planted may
be forced to travel further for grazing resources (particularly during the
growing season when grazing in harvested fields is not possible). Finally,
Sukuma herders also report that large herds move more quickly through
patches and it is more difficult to get a large herd to settle down in a single
area for grazing. Thus, the size of a household’s herd may affect their
herding radius by forcing the herder go farther to find sufficiently large
grazing patches.
The herding radius circumscribes the area in which grazing takes place,
but additional information is necessary for characterizing the distribution
of grazing within its bounds. Grazing may be spread evenly within the
534
Coppolillo
herding radius, concentrated around home or at the edges of the herding
radius, or skewed toward or away from particular landscape features. Obviously, the choice of locally optimal vegetation communities will affect this
distribution in site-specific ways. More generally however, the distributions
of water and other cattle-keeping settlements are likely to affect the directions herds travel from home and consequently, the landscape-scale distribution of grazing. The distribution of grazing can be skewed toward or
away from water points and other settlements. The ecological significance
of these shifts in grazing may depend on whether they are consistent
throughout the year or occur during certain seasons. For example, grazing
pressure around pastoral settlements may be homogenous when viewed
over an entire year, but the timing of grazing may differ for areas in different
directions relative to water. Since timing of grazing may be as important
as the overall stocking rate (Westoby et al., 1989; Walker, 1993), the distribution of water is likely to affect the types and distribution of impacts.
The final component of the model concerns livestock productivity. Of
the household-level factors listed above, I examine the effects of herd size
and settlement density on four measures of productivity. I also ask whether
traveling farther (i.e., increasing herding radius) affects cattle productivity.
Productivity can be assessed using a variety of measures which vary in
their degree of aggregation. Intake rates and foraging behavior provide
immediate measures of productivity, while milk yields and body condition
give an aggregate picture of productivity on weekly to monthly time scales.
The conceptual model developed above provides a framework for
examining how household- and landscape-level factors shape the distribution of grazing in pastoral and agropastoral areas. Below, I describe the data
collected and analytical procedures used to evaluate the factors affecting
herding practices and the resultant distribution of grazing.
DATA COLLECTION
I constructed a sample of 24 Sukuma households spread throughout
the study area (Fig. 1) and stratified to capture a range of herd sizes (as a
measure of wealth). ‘‘Household’’ is defined here as an individual herding
unit (similar to Dahl & Hjort, 1976). In cases where families cultivate
separately but herd their animals together I have treated them as a single household.
To link herding practices to household economic conditions, I conducted demographic surveys in all 24 households. Surveys covered the
number of people living in the household, their land and livestock holdings
and cultivation practices. Based on these data I calculated three related
Landscape Ecology of Herding
535
measures of labor stress in each household. I defined ‘‘total labor’’ as the
number of individuals in a household over the age of 6. This age was chosen
as the cutoff because it is around six that children are able to start herding
and weeding, and because it is consistent with other studies examining
labor in pastoral societies (Roberts, 1996; Sieff, 1997). Despite the widely
held view that herding is a purely male activity, females did herd in this
and other studies (Fratkin, 1991; Sieff, 1997). Also, during the wet season
everyone in the household contributes to weeding crops, and it is only after
each day’s weeding that herding can begin. Therefore, the total number of
people in the household, not just the males, is the relevant measure of
labor as it relates to herding. ‘‘Labor demand’’ is defined as total labor in
the household divided by the total area cultivated by the household. The
final measure was simply the total area cultivated by the household (all
crops were weighted equally).
To quantify spatial patterns of pastoral land use I recorded the movements of cattle herds from the 24 focal households on 73 full-day herd
follows (similar to Coppock et al., 1986a, 1986b; Homewood and Rodgers,
1991; Nyerges, 1982). I followed the primary (adult) cattle herd from the
time they left their enclosure in the morning until they returned in the
evening. Like other pastoral groups, Sukuma herd their animals separately
based on age, species, and sometimes reproductive status (e.g., see Coppock
et al., 1986a; Homewood & Rodgers, 1991; McCabe, 1985, 1990a; Nyerges,
1982; Sieff, 1997; Turner, 1999). Because the primary herds constitute the
majority of livestock biomass and because calves, goats, and sheep are
often left unherded, I followed only primary cattle herds. I followed each
household’s herd during the dry season (Sep.–Dec., 1995), the early part
of the wet season (Jan.–early Mar., 1996) and the late part of the wet
season (late Mar.–May, 1996). During each herd follow, I recorded herd
locations at least every hour (more often when herds were moving quickly
or changing directions frequently) using a hand-held global positioning
system (GPS). I refer to these data below as ‘‘route locations.’’2
Five times per hour, I observed a randomly chosen animal’s behavior
during a three-minute focal observation. During each observation the focal
animal’s bites, steps, and chews were recorded continuously. At the end
of the three minutes the mean bite size was visually estimated on a scale
from 1 to 5, so that an overall intake rate (bites * bite-size) over three
minutes could be calculated. Through the course of the study, 2,146 observa2
The GPS data were not spatially corrected since the error introduced (20–50 m) is smaller
than the spread of a typical herd and not significant relative to the distances covered during
a day’s herding.
536
Coppolillo
tions were recorded yielding over 268 hours of observation time. Behavioral
records for the entire study are plotted in Fig. 2.
These data, referred to hereafter as ‘‘behavioral records,’’ serve two
purposes. First, spatially referencing behavioral records (see Data Analysis)
makes it possible to analyze the distribution of grazing around pastoral
settlements and evaluate the factors affecting this distribution. Second,
intake rates and foraging behavior provide immediate measures of energy
intake (and thus resultant livestock productivity) for herds subjected to
different herding practices and for discrete points on the landscape.
Every 20 min I scanned the entire herd (or the portion of the herd that
was visible) and recorded each individual’s behavior as feeding, walking,
standing, ruminating, lying, or drinking. As in other pastoral studies (Nyerges, 1982) these ‘‘scan samples’’ provide a quantitative measure of the
entire herd’s activity, as opposed to the behavioral records, which provided
more detailed data on a single individual’s behavior (Altmann, 1974). During the study, the behavior of 57,532 individuals was recorded in 1,341
scan samples.
The locations of all cattle-keeping households in the study area (N ⫽
150) were recorded using the GPS. These data were used to calculate
Fig. 2. The locations of all behavioral records. Legend as in Fig. 1.
Landscape Ecology of Herding
537
settlement densities (see Data Analysis, below). As with the focal sample,
households were identified based on cattle herding units, rather than cultivation or marriage units.
I used three measures of cattle productivity. First, I observed individual
animals foraging behavior and calculated intake rates (described above)
to provide an immediate (least aggregated) measure of productivity. The
importance of walking for pastoral animals’ energetics and productivity has
been demonstrated elsewhere (Coppock et al., 1986a, b; Homewood and
Rodgers, 1991; but see Western and Finch, 1986). To monitor walking (and
associated energetic costs), I counted the total number of steps taken by
the animal and the number of ‘‘walking bouts,’’ defined as ten steps taken
without taking a bite. A second, more aggregated measure of productivity
is milk yield, which reflects an animal’s diet over the last few days to a
week. Yields were measured from 532 animals (in the wet seasons only)
by weighing the milk from each lactating female before it was placed into
a common vessel (similar to Homewood and Rodgers, 1991). Note that this
records the amount taken for human consumption, not the total produced.
Finally, I used body condition as the most aggregated measure of herd
productivity. Body conditions of 1,725 animals were scored on a nine-point
scale developed specifically for zebu cattle (Nicholson and Butterworth,
1986). The scale quantifies animals’ body condition based on the visibility
of an individual’s ribs and vertebrae, and muscle mass. Reference cards
with photographs of animals exemplifying each condition score were used to
avoid seasonal and inter-herd biases. Condition scores are tightly correlated
with animals’ overall weight and reflect the animal’s diet and nutrition over
the last few weeks or months (Nicholson and Butterworth, 1986). For each
herd I scored the body condition of 30 individuals. For herds smaller than
30, I scored the entire herd. Lactating females’ body condition was recorded
during milking. Preliminary data collection showed that herding practices
were consistent over one month intervals, so a single day’s herding is
generally representative of the recent past, validating its association with
aggregated measures of productivity.
DATA ANALYSIS
I plotted each behavioral record spatially by interpolating between the
previous and subsequent route locations. This assumes that herds move at
a constant pace and in a straight line between each route location, both
reasonable assumptions given the frequency of records and the fact that
when a herd’s pattern of movement changed additional route locations were
recorded. This allowed me to spatially reference each behavioral record.
538
Coppolillo
All behavioral records were then entered into a Geographic Information System (ArcInfo, 1997, henceforth GIS) as text files. Each behavioral
record was referenced according to its distance and compass bearing from
home. These data make it possible to quantitatively examine the total
distance traveled, the herding radius, and the spatial distribution of herding
records (as a measure of grazing intensity) around each settlement.
The calculation described above characterizes herds’ locations and
movements relative to grid north. A more relevant expression of direction
for testing hypotheses relating to water distributions may be ‘‘toward or
away’’ from water. In order to compare the movements of each household’s
herd relative to its home and dry season water source, I created a standardized grid plotting all behavioral records relative to these two points. Figure
3 is a schematic diagram of this process. Conceptually, this process is analogous to plotting each herd’s movements on a small map and then aligning
the maps for all herds so that water falls due ‘‘north’’ of each settlement
(see the lower panel of Fig. 3). This effectively standardized every herd’s
movements allowing them to be compared and analyzed collectively.
To achieve this conversion quantitatively, I used the GIS to calculate
the direction from each household to its dry season water source. The
directions of all behavioral records associated with that household’s herd
were then adjusted relative to this value. Thus, the value given to each
behavioral record’s direction is the difference between the bearing from
the record itself to home and the bearing from that household’s dry season
water source to home. All behavioral records were then given coordinates
that expressed location relative to home (which was at the origin on an
imaginary grid) and the dry season water source (lying due ‘‘north’’ of the
homestead on the imaginary grid).
I statistically analyzed the spatial pattern of grazing in relation to water
by comparing the numerical distribution of directions for behavioral records
(expressed in degrees from 0–180) with a uniform distribution using a
Kolmogorov Smirnov Z statistic (I performed all statistical analyses in SPSS,
version 7.5, 1995). I used a uniform distribution because it corresponds to
the null hypothesis that the spatial distribution of behavioral records (i.e.,
grazing) is random with respect to water.
For each herding day, I defined the herding radius as the maximum
distance from home at which a herd was observed during that day (see the
top panel of Fig. 3). The factors affecting herding radius (distance to water,
labor, herd size, and the density of other settlements) were analyzed using
linear regression.
To calculate settlement densities I entered the coordinates of all cattlekeeping households in the study area (N ⫽ 150) into the GIS as text files.
Next, the entire area within 9 km of the 24 focal households was broken
Fig. 3. An illustration of the data collected for each herding day (above) and its manipulation
in order to compare different households’ movements relative to water (below). See text
for details.
540
Coppolillo
up into 250 meter grid cells. I chose 9 km as a boundary because this was
the maximum distance from home herds were observed to travel, so it is
a conservative estimate of the entire area available to pastoral herds. I then
used the GIS to calculate the density of settlements within two kilometers
of each grid cell. This calculation used all 150 settlements in the study area,
not just the 24 focal households. Because the majority of grazing occurred
within two kilometers of home, this distance includes an ecologically relevant portion of the landscape without obscuring local heterogeneity.
I analyzed the spatial distribution of grazing relative to settlement
densities by comparing the settlement densities in areas used by herds to
those available to them. If herders show a preference for high or low
settlement densities, these distributions will differ significantly. Under the
null hypothesis, herders will show no preference, and use will occur in
proportion to what is available. Preference values for each of 21 equal
classes of settlement densities were calculated as (O ⫺ E)/(O ⫹ E) where
O is the proportion of behavioral records in each class, and E is the proportion of area occupied by that class in the entire area available to herds.
This index ranges from ⫺1 to 1, indicating strong avoidance and strong
preference, respectively.
RESULTS AND DISCUSSION
I divide this section into two parts. First, I report the household-level
factors affecting the herding radius (i.e., how far herds travel from home)
and the distribution of grazing within the herding radius. In the second
section I report the different measures of productivity resulting from different herding practices. Specifically, I examine how traveling farther from
home (herding radius), keeping larger herds and living in or using densely
settled areas affect herd productivity (intake rates and foraging behavior,
milk yields, and body condition).
What Affects How Far Herds Range from Home?
As mentioned earlier, Sukuma settlements are not always in close
proximity to water; households’ distances to perennial water ranged from
just over 500 meters to just under four kilometers. Distance from the
household to water was the strongest predictor of the dry season herding
Landscape Ecology of Herding
541
radius (Fig. 4 and Table I). The fact that most points in Fig. 4 cluster along
the line of x ⫽ y implies that during the dry season herds mostly travel
only as far as their water source. Recall that the herding radius is ecologically
significant because it defines the overall area affected by grazing as well
as the extent of overlap between nearby households. Therefore, water
availability is an important determinant of the landscape-scale distribution
of grazing. This portion of the analysis was carried out only for dry season
herding days, because water was available virtually everywhere during
the wet season. Most studies of water availability have been in pastoral
ecosystems with far less rainfall (c. 100–200 mm p.a.; see Knight, 1995;
McCabe, 1990b; Nyerges, 1982; Verlinden, 1997; Western, 1975). The results
presented here suggest that even in a relatively wet area receiving about
five times the rainfall of more arid pastoral systems, the distribution of
water influences land use and therefore is critical in modeling or assessing
the impacts of pastoral grazing.
Sukuma herders report that the size of a herd affects its requirements
and how it is herded, so I examined whether the effect of herd size is
apparent in landscape-scale daily herding practices. Herding radius in the
dry and early wet seasons and when all seasons were considered together
was (positively) affected by the size of a household’s herd. This effect could
be spurious if the owners of larger herds lived farther from water, but this
Fig. 4. Herding radius vs. each household’s distance to water during the late dry season.
542
Coppolillo
Table I. Factors Affecting Herding Radius and Leaving Timesa
Season
r2
Herding radius
Dry
0.814
⬍0.005
Herding radius
Whole year
Dry
Dry (mult. reg.)
Early Wet
Late Wet
0.177
0.522
t ⫽0.717
0.483
0.086
⬍0.005
⬍0.005
n.s.
⬍0.005
n.s.
Herding radius
Dry
Early Wet
Late Wet
0.0
0.002
0.0
n.s.
n.s.
n.s.
Leaving time
Dry
Early Wet
Late Wet
0.004
0.0
0.013
n.s.
n.s.
n.s.
Herding radius
Dry
Early Wet
Late Wet
0.029
0.016
0.056
n.s.
n.s.
n.s.
Leaving time
Dry
Early Wet
Late Wet
0.013
0.001
0.0
n.s.
n.s.
n.s.
Herding radius
Dry
Early Wet
Late Wet
0.023
0.006
0.036
n.s.
n.s.
n.s.
Leaving time
Dry
Early Wet
Late Wet
0.002
0.012
0.002
n.s.
n.s.
n.s.
All Seasons
Dry
Early Wet
Late Wet
0.12
0.001
0.084
0.170
n.s.
n.s.
n.s.
n.s.
Predictor variable
Response variable
Distance to water
Herd size
Labor
Labor demand
Area cultivated
Settlement density
Herding radius
P value
Significance is reported when P ⬍ 0.05.
a
was not the case since herd sizes, which ranged from 12 to 224 animals,
were not correlated with a household’s distance to water (r2 ⫽ 0.0019,
P ⫽ 0.843). However, in the dry season if herd size is analyzed as part of
a multiple regression including distance to water, its effect was no longer
statistically significant. This is understandable given the strong effect that
the distance to water has on dry season herding radius. The reasons why
larger herds travel farther are complex, involving herd productivity and
the foraging behavior of individuals in large and small herds. For this reason
a thorough discussion of the effects of herd size is presented after the
productivity data.
Another factor potentially affecting how far herds travel from home
is the amount of labor a household can devote to herding. This may be
simply a function of the number of people in the household or of the
competing labor demands from other aspects of household production. In
Landscape Ecology of Herding
543
the Sukuma case, cultivation may substantially drain labor from herding.
Therefore, I investigated the effects of total household labor, the amount
of land cultivated by the household and a ratio of the two. I examined
whether these independent variables affected the herding radius and the
times herds were let out to be herded (since weeding and cultivating is
generally undertaken in the early morning). The results, summarized in
Table I, show that none of the labor or cultivation variables had a measurable effect on daily herding practices.
The lack of relationships between labor, cultivation, and herding is
surprising because cultivation is a contentious topic in policy debates surrounding pastoral production and conservation (e.g., see McCabe et al.,
1992), and given the well-recognized time constraints on herding (Bayer,
1990; Homewood and Rodgers, 1991), it seems logical that a time-consuming enterprise like cultivation would drastically change herding practices.
Because all households in the study cultivate, these data cannot shed light
on whether the initial adoption of cultivation qualitatively changes pastoral
land use. However, these data do suggest that once cultivation has been
undertaken, increasing or decreasing the area cultivated or the number of
individuals participating has a minimal effect on herding practices. Since
many households cultivate, even in groups once characterized as ‘‘pure
pastoralists’’ (Brockington and Homewood, 1999; McCabe et al., 1992;
McCorkle, 1992; Dyson-Hudson and Dyson-Hudson, 1980), this result may
have wider generality than is initially apparent.
The discussion above does not imply that labor is unimportant in
pastoral households, only that it does not strongly affect daily herding
practices. Labor is likely to be very important in the non-herding aspects of
pastoral production, including milking, processing animal products, treating
sick animals, marketing pastoral products, or in non-pastoral activities such
as cultivation (Morton, 1990; Niamir, 1990; Fratkin and Smith, 1995). Amanor (1995) cites the importance of labor for herding in pastoral systems
where seeds are collected or fodder is cut for livestock, but this assertion
is not tested quantitatively (see Sieff, 1997; and Turner, 1999, for more on
herding and labor). And cultivation does, of course, drastically change
landscape structure in ways that certainly affect herding, so a lack of association with herding radii or the times herds leave home is obviously not an
indication that cultivation is ecologically insignificant.
A final factor that could affect the distances herds range from home
is the density of other cattle-keeping settlements around a household. More
settlements (and more cattle) could deplete grazing resources and force
herds to travel farther to find suitable forage. Focal households were situated in areas with densities spanning an order of magnitude, from .36 to
3.6 settlements km⫺2, but this variation did not affect the herding radius in
any single season or when the whole year is lumped (Table I). Settlement
544
Coppolillo
densities did however, affect the spatial distribution of grazing within the
herding radius, so their importance is discussed below.
What Affects Patterns of Grazing Within the Herding Radius?
With the total area used by herds circumscribed by the herding radius,
it is then necessary to examine the distribution of herding within that area.
Two factors were considered here: the distribution of dry-season water and
the distribution of cattle-keeping settlements. The distribution of grazing
was skewed toward perennial water in the dry season and away from it in
the wet season. Figure 5 shows the observed dry and wet season distributions
of use (standardized relative to water).
Fig. 5. Seasonal distributions (expressed in degrees) of herd locations relative to water. Dry
and Wet season distributions are shown on the left and right respectively. Histograms (top)
represent the number of behavioral records observed in each direction class. Directions are
calculated for each behavioral observation according to its direction from home relative to
that household’s dry season water source (see text for details). Detrended Q–Q plots (bottom)
show seasonal distributions plotted against a uniform distribution. Under uniform distributions
of grazing relative to water, all points would fall on the horizontal lines. Kolmogorov-Smirnov
Z statistics, based on the maximum deviation from uniform (on the vertical axes), confirm
the seasonal differences shown in the histograms (dry: K–S Z ⫽ 8.673, P ⬍ 0.0005; wet: K–S
Z ⫽ 6.219, P ⬍ 0.0005)
Landscape Ecology of Herding
545
The seasonal shifts in grazing, taken together with the strong effect of
water availability on herding radius, suggest that even in a relatively wet
area like the Rukwa Valley, water is an important factor structuring the
livestock grazing system. The results presented here also support Western’s
(1975) assertion that manipulating the distribution of water can influence
pastoral land use systems. Again, this is noteworthy given the relatively
greater rainfall in the Rukwa Valley. Clearly, water is a powerful but
potentially dangerous management tool. Restricting access to water has
been used to alleviate perceived resource conflicts between wildlife and
livestock and minimize grazing in sensitive areas, but it has also led to
intense conflicts between pastoralists and protected area managers (see
reviews in Western, 1994; Lindsay, 1989). Another option may be to provide
alternate water sources, but this can have serious negative consequences
as well [Knight, 1995; Verlinden, 1997; but see Burkett and Thompson
(1994) and Thrash (1998)]. McCabe (1990b) and others have described the
social institutions that influence land use by regulating access to water.
Water development should be pursued with care; an understanding of
the social context in which access to water is regulated is critical to any
management action.
An important caveat is that these results should not be interpreted as
demonstrating a ‘‘piosphere effect’’ (defined as heavily grazed and often
degraded area surrounding water points see Sinclair and Fryxell, 1985;
Thrash, 1998). Recall that the animals in this study are herded and must
be returned to their settlements each night. This constrains their ability to
stay close to water, as unherded livestock and wild ungulates do (Estes,
1991; Lamprey, 1964; Western, 1975) and may be the reason why other
pastoral systems tend not to show piosphere-type effects [Roe, 1984; and
see Coppolillo (in press) for a more detailed discussion of the landscapelevel grazing intensity when many households are aggregated].
Settlement density also affected the spatial distribution of herding, but
perhaps surprisingly, herders’ use was strongly skewed toward high and
away from low density areas throughout the year (Fig. 6).3 Seasonal distributions were not markedly different from the entire year, so a single analysis
was performed. Preference values calculated for each class of settlement
densities were strongly related to the lower limit of each density class
(Fig. 7).
The year-round preference for high density areas is probably due to
two factors: the integration of herding and cultivation and extremely high
wet season productivity in undisturbed areas. During the dry season cattle
are often grazed in harvested agricultural fields, which are all in higher
3
Eliminating the lowest density class does not change the results observed.
Fig. 6. Settlement densities within 9 km of households (right) and settlement densities actually used by households (left). Densities are
the number of settlements per square kilometer. Used and available were compared using a chi-squared statistic, which confirmed their
difference (␹2 ⫽ 7541.37, P ⬍ 0.0005).
546
Coppolillo
Landscape Ecology of Herding
547
Fig. 7. Relationship between maximum settlement density in each class and the preference
value for that class. Trend line is a quadratic best fit.
density areas. In general, herd owners graze their animals in their own
cultivated areas (and can under traditional land tenure exclude all others
from grazing in areas they have cultivated) but occasionally they pay Pimbwe horticulturalists for access to their fields as well. Wet season preference
for more densely settled areas probably results from extremely high productivity which produces very tall and fibrous grasses. In some areas Themeda
triadra, normally a very palatable grass, was over three meters tall during
the wet season and consequently of little use as forage. Sukuma herders
appear to be using high density (and often disturbed) areas as grazing lawns
where shorter, more palatable forage can be found. Birley (1982) reports
similar use of high density areas to keep standing vegetation low and reduce
ectoparasite loads.
How Do Herding Practices Affect Cattle Productivity?
This section examines how variation in herding practices affects measures of herd productivity. Three primary questions are addressed: how
does traveling farther from home, keeping cattle in large herds, and using
areas of varying settlement density affect herd productivity? The measures
548
Coppolillo
of productivity used are herds’ mean intake rates, foraging behavior, milk
yields, and body conditions.
First, how does traveling farther from home affect herd productivity?
Herding radius did not affect mean daily intake rates during any single
season or over all seasons (Table II). Looking at individual behavioral
records (not aggregated for entire days), distance from home had a significant negative effect on intake rates only in the late wet season (Table II),
implying that being further from home was associated with lower intake
rates. However, the effect only explained around 1% of the variance during
the late wet season, so I conclude that traveling farther from home did not
affect intake rates in a biologically meaningful way.
It is possible that more palatable forage is available farther from Sukuma settlements. This analysis would not resolve that difference because
intake rates were calculated without regard to the species being eaten.
However, if this was the case, herds that traveled farther would have similar
intake rates but receive a more nutritious diet. Consequently, one would
expect higher milk yields and body conditions for herds that traveled farther,
but these were not observed. In fact, herding radius had no effect on body
condition and a negative effect on milk yields (Table II). This effect was
strong, with lactating females in herds that ranged farthest from home
producing around one third of the milk produced by lactating females in
herds that showed the smallest herding radii. This is not surprising given
the fact that lactating ungulates are energetically stressed (Clutton-Brock
et al., 1982). Under the current circumstances it seems that traveling further
provides little nutritional benefit, and energetic costs are responsible for
Table II. Seasonal Effects of Herding Radius on Productivitya
Season
r2
Mean daily intake
All
Dry
Early wet
Late wet
0.029
0.144
0.020
0.136
n.s.
n.s.
n.s.
n.s.
Individual intake
Dry
Early wet
Late wet
0.010
0.001
0.012
n.s.
n.s.
0.004
Milk yields
Early wet
Late wet
0.032
0.451
n.s.
0.001
Herd’s body condition
Dry
Early wet
Late wet
0.071
0.133
0.047
n.s.
n.s.
n.s.
Lactating females’ body condition
Early wet
Late wet
0.088
0.183
n.s.
n.s.
Response variable
Significance is reported when P ⬍ 0.05.
a
P value
Landscape Ecology of Herding
549
decreased milk production from lactating females. With increasing cattle
densities, a decrease in range condition could eventually tip the balance in
favor of traveling further for higher quality forage, but it appears that this
is not yet the case in the Rukwa Valley. Analysis of vegetation across a
gradient of cattle densities should help resolve this question and is currently underway.
Keeping cattle in larger herds had complex effects on productivity.
Herd size did not have a significant effect on intake rates in any season
when considering either mean daily intake rates or individual intake rates
during behavioral records (Table III). However, herd size substantially
affected foraging behavior. In the behavioral records, herd size was positively related to daily mean stepping rates over the whole wet season (Table
Table III. Effects of Herd Size on Productivitya
Season
r2
Mean daily intake
All
Dry
Early wet
Late wet
0.039
0.043
0.070
0.001
n.s.
n.s.
n.s.
n.s.
Individual intake
Dry
Early wet
Late wet
0.011
0.005
0.001
n.s.
n.s.
n.s.
Mean daily stepping rate
Dry
Early wet
Late wet
0.003
0.672
0.311
n.s.
⬍0.0005
0.006
Steps during feeding bouts
Dry
Early wet
Late wet
0.066
0.294
0.213
n.s.
0.007
0.027
Steps during walking bouts
Dry
Early wet
Late wet
0.004
0.601
0.249
n.s.
⬍0.0005
0.015
Number of walking bouts
Dry
Early wet
Late wet
0.008
0.641
0.361
n.s.
⬍0.0005
0.002
P Animals Feeding
All
0.064
0.039
Response variable
P value
P Animals Walking
All
0.077
0.023
Milk yields
Early wet
Late wet
0.033
0.014
n.s.
n.s.
Herd’s body condition
Dry
Early wet
Late wet
Whole wet
0.042
0.197
0.060
0.017
n.s.
0.034
n.s.
n.s.
Lactating females’ body condition
Whole wet
0.005
n.s.
Significance is reported when P ⬍ 0.05.
a
550
Coppolillo
III). Herd size was also positively related to the mean number of walking
bouts (the number of times individual animals took greater than 10 steps
without a bite) over the whole wet season.
Neither the stepping rate nor the mean number of walking bouts were
related to herd size during the dry season (Table III). This lack of association
is not simply because all herds were traveling to water; this portion of the
analysis excludes transit to and from water. It is worth noting that for small
herds, stepping rates and the number of walking bouts were highest during
the dry season. In fact, the lack of association in the dry season is explained
by the fact that small herds’ stepping rates and walking bouts were as high
as those of large herds. In other words, during the wet season it pays
(through decreased walking) to keep animals in smaller herds, but during
the dry season, all herds have high stepping rates.4 This suggests an interaction between seasonal resource availability and herd size.
To summarize, herd size did not affect intake rate but large herds
spent more time walking and less time feeding. It follows logically to ask
why large herds walk more. Increased walking seems to stem from the fact
that within large herds there are always a few individuals not actively
feeding but walking from one feeding station (sensu Bailey et al., 1996) to
the next, leaving other animals behind. Because individuals tend to stay
within a certain part of the herd (i.e., the front, middle, or back), they are
often forced to stop feeding and walk to their usual position. These are
the ‘‘walking bouts’’ reported above. This leap-frog behavior by individuals
within the herd keeps large herds constantly moving. As mentioned earlier,
increased walking is likely to affect animals’ overall energy balance, and if
the energetic costs of walking are not offset by greater resource availability
in areas farther from home, it is likely responsible for the lower body condition of larger herds and lower productivity from herds that travel farther.
The increased walking also helps to explain why large herds traveled
farther in the late dry and early wet seasons. The process described above
precludes large herds from using smaller high quality patches close to home
because the herd simply moves through them too quickly or they provide
an insufficient number of feeding stations to occupy the whole herd at once.
The result is that large herds end up using larger patches farther from home.
Given the differences in feeding behavior one might expect to see
differences in milk yields associated with herd size, but herd size did not
4
Breaking the overall stepping rate down into steps taken during walking bouts and steps
taken during feeding bouts shows that herd size affected both stepping rates throughout the
wet season and had no effect on either stepping rate during the dry season (Table III). These
results were corroborated by the scan samples, which showed that across all seasons larger
herds had a significantly smaller proportion of animals feeding (around 10% fewer) and a
significantly larger proportion of animals walking (about 15% more).
Landscape Ecology of Herding
551
affect milk yield during the early or late wet seasons (Table III). When
analyzed across the whole wet season, herd size was not associated with
lactating females’ or the whole herd’s body condition. However, looking
at individual seasons, herd size did have a significant negative effect on the
whole herd’s body condition during the early wet season (Table III). This
is a critical time for many pastoral animals since their body condition is
still low from the dry season and they must endure cold and rain before
realizing the benefits of better forage.
The increased traveling of large groups and productivity costs of travel
are both consistent with other studies. Homewood and Rodgers (1991)
found that milk yields dropped off sharply as the distance herds traveled
increased. In primates, Altmann et al. (1993) found that baboons’ body
condition was lower in groups that ranged further, and Isbell (1991) and
Isbell et al. (1998) report that larger groups of patas and vervet monkeys
ranged further per unit time. The negative relationship between herding
radius and body condition presented here contrasts with the findings of
Western and Finch (1986), who report that steers were able to mitigate the
effects of increased traveling by lowering their metabolism. A probable
explanation for the difference between their results and those presented
here is that Western and Finch (1986) focus only on steers. It is unlikely
that other cattle, particularly energetically stressed lactating females, would
be able to respond to increased travel in a similar manner.
Another consideration is that large herds’ body condition was lower
in the early wet season, which is the time when Sukuma herders report
that disease takes the most severe toll on their animals. Taken together,
the lack of resource scarcity and extensive use of high density areas suggest
that disease may also be contributing to the decreased body condition in
large herds. Rawlings et al. (1994) and Watcher et al. (1993) have shown
that land use and livestock densities can affect the densities of tsetse flies.
Grootenhuis and Olubayo (1993) also report increases in ectoparasite loads
for wildlife living in association with livestock. Since each herd returns to
the same enclosure and watering hole each day, larger herds may encounter
a more substantial challenge from parasites. Future research should focus
on disentangling the effects of herd size, herding practices, and settlement
densities on disease interactions among pastoral herds.
The herd size effects reported here also cast light on the practice of
herd splitting. Other pastoralists, particularly those living in arid environments and relying heavily on their herds for subsistence, split herds more
extensively by species and even reproductive status (Coppock et al., 1986a;
McCabe, 1994; Nyerges, 1982; Sieff, 1997). Herd splitting allows them to
tailor herding to specific species’ needs (e.g., Coppock et al., 1986b), and
take advantage of differences in species’ feeding ecology and behavior (e.g.,
552
Coppolillo
Nyerges, 1982). The data presented here suggest that even within singlespecies herds, limiting group size appears to help herders reduce the distance their animals walk while foraging.
A final question is whether living in and/or using densely-settled areas
affects herd productivity. The density of other settlements around households did not affect mean daily intake rates when examined over the entire
year (Table IV). When looking at only the dry season, settlement density
did have a negative effect on mean daily intake rate, but this relationship
did not hold for the wet seasons. A related question is whether traveling
to areas with lower settlement densities provides higher intake rates. To
answer this question I examined the relationship between intake rate during
individual behavioral records and the settlement density at the location of
each record. There were no significant relationships. Further, the settlement
density around focal households did not affect milk yields (Table III).
When looking at the entire herd’s body condition, settlement density had
a significant positive effect during the dry season but not the wet seasons.
Lactating females’ body condition was positively related to settlement density during the early wet season and unrelated to settlement density during
the late wet season (Table III).
Two apparently inconsistent results were that dry season daily intake
rates were negatively related to households’ settlement density, but spatial
analysis of the individual behavioral records showed no relationship between settlement densities and intake rates. Furthermore, use was strongly
skewed toward higher settlement densities throughout the year, including
the dry season, and body condition was positively related to settlement
Table IV. Effects of Settlement Density Productivitya
Season
r2
Mean daily intake
All
Dry
Early wet
Late wet
0.003
0.834
0.042
0.002
n.s.
0.004
n.s.
n.s.
Individual intake
Dry
Early wet
Late wet
0.008
0.000
0.004
n.s.
n.s.
n.s.
Milk yields
Early wet
Late wet
0.051
0.099
n.s.
n.s.
Herd’s body condition
Dry
Early wet
Late wet
0.189
0.028
0.006
0.030
n.s.
n.s.
Lactating females’ body condition
Early wet
Late wet
0.199
0.042
0.049
n.s.
Response variable
Significance is reported when P ⬍ 0.05.
a
P value
Landscape Ecology of Herding
553
density during the early wet and dry seasons. Reconciling these results may
be possible in light of two observations: first, during the dry season all
cattle were herded from sunrise to sunset, implying that they were time
limited; and second, high density areas tend to be nearer to water. Proximity
to water may give herds from densely settled areas more time to feed
because travel time to and from water is shorter. Consequently, herds living
at high densities could feed less intensely (obtaining lower mean intake
rates) over a longer grazing time as they travel less than herds living at lower
settlement densities. Bayer (1990) reports that time constrained pastoral
animals fed more consistently and intensely than unconstrained animals to
maintain similar levels of productivity. Thus, herds can experience lower
mean intake rates and still have better body condition during the dry season.
SUMMARY AND CONCLUSIONS
The objectives of this study were to examine the factors affecting the
spatial patterns of daily herding among Sukuma agropastoralists living in
the Rukwa Valley, and to assess how different land use practices map onto
measures of herd productivity. The results, summarized schematically in
Fig. 8, demonstrate considerable household-level variation in herding practices and in resultant levels of cattle productivity. The data presented here
support the contention that individual-level analyses reveal significant variation not apparent at the population level.
In the Rukwa Valley, the distribution of dry season water structured
the landscape-scale distribution of grazing throughout the year, not just
during the dry season. Water availability strongly affected the distances
herds ranged from home in the dry season and the distribution of grazing
around pastoral settlements. Since the timing of grazing may be as important
as herbivore density (Westoby et al., 1989; Walker, 1993), the location of
water may affect rangeland vegetation both near and far from water sources.
A second noteworthy result is that herders showed strong preferences
for densely settled areas relative to what is available. This has a number
of implications. First, it highlights the spatial aggregation of livestock and
agrarian production. Use of high density areas reflects herders’ choice of
harvested and fallow agricultural fields for grazing. This means that livestock
production may indirectly affect landscape structure since the amount of
land under cultivation and in fallow, the ratio of the two, and their spatial
configurations may reflect herding as well as agrarian considerations. Herders’ preference for high density areas also makes encroachment into surrounding, relatively undisturbed areas unlikely (provided, of course, that
these areas remain unsettled and uncultivated). Both landscape structure
and the potential for encroachment are important aspects of pastoral sys-
554
Coppolillo
Fig. 8. Diagrammatic summary of household-level factors affecting herding practices (top)
and land use and household-level factors affecting herd productivity (bottom).
Landscape Ecology of Herding
555
tems for biological conservation, given the fact that pastoral and agropastoral peoples live adjacent to or surrounding nearly all of East Africa’s
national parks and game reserves (Enghoff, 1990).
The relationships between herding practices and productivity were
more complex. Individuals from larger herds were observed to walk more
while actively foraging and engage in more walking bouts (taking ten steps
without taking a bite). However, the effects of large herds’ increased walking were not apparent in intake rates or milk yields. The increased walking
of large herds may explain why they range farther from home and highlight
the importance of herd splitting.
To what extent can these results be applied to other pastoral systems?
Interestingly, similar shifts in grazing toward water in the dry season and
away in the wet season were documented by Western (1975) in an arid
ecosystem (Kajiado District near Amboseli N.P., Kenya). Since the Rukwa
Valley receives nearly five times the rainfall of the Amboseli Basin, it is
possible that this is a generalized phenomenon which should hold for most
or all pastoral systems. One exception may be when herders alternate
forage and watering days and travel in different directions on each day
(Western and Finch, 1986; Homewood and Rodgers, 1991). The ubiquity
of herd splitting among other pastoral groups suggests that the effects of
herd size observed in this study are also important in other pastoral systems.
All pastoral groups are essentially central-place foragers when considered at a daily time scale because they come and go from the same homestead each night. The analytical methods presented here can be applied to
any pastoral herding situation given a known daily starting point. In fact,
this method could be applied to any number of resource systems where
impacts are distributed around a focal point. Applying this method comparatively to other pastoral groups will provide a much greater level of generality and a broader understanding of pastoral systems. Longer term studies
capturing inter-annual variation through drought and wet years and experimental approaches will undoubtedly enhance our ability to recognize the
mechanisms driving herding systems.
ACKNOWLEDGMENTS
I extend my deepest thanks to C. G. Coppolillo for encouragement
and assistance in data collection during what was often difficult fieldwork.
I also thank M. Borgerhoff Mulder for extensive guidance and sustained
enthusiasm, T. P. Young, R. E. Plant, T. M. Palmer, L. M. Ruttan, and T.
M. Caro for comments, useful discussions and logistical cooperation, P. W.
Grant for technical assistance with GIS, N. Willets for statistical consulting
556
Coppolillo
and three anonymous reviewers for comments which substantially improved
the manuscript. I am grateful to A. O. Msago, E. E. Magambo, F. Mwanansoro, F. Bashumika and D. Mwakasungura for introductions and institutional support, and to Tanzania National Parks, The Tanzania Commission
for Science and Technology, and The Serengeti Wildlife Research Institute
for research permission. I am also deeply indebted to the Sukuma of Kibaoni, Manga, and Mirumba who fed and housed me, answered endless
questions and immeasurably improved my research by contributing countless insights and advice. This research was funded by the U.S. Agency for
International Development Small Ruminant CRSP Program, a National
Science Foundation Predoctoral Fellowship, and Friends of Conservation.
REFERENCES
Altmann, J. (1974). Observational study of behavior: Sampling methods. Behaviour 49:
227–267.
Altmann, J., Schoeller, D., Altmann, S. A., Muruthi, P., and Sapolsky, R. (1993). Body size
and fatness of free-living baboons reflect food availability and activity levels. American
Journal of Primatology 30: 149–161.
Amanor, K. S. (1995). Dynamics of herd structures and herding strategies in West Africa: A
study of market integration and ecological adaptation. Africa 65: 351–393.
Århem, K. (1985). Pastoral Man in the Garden of Eden: The Maasai of the Ngorongoro
Conservation Area, Tanzania, Research Reports in Cultural Anthropology, University of
Uppsala, Uppsala, Sweden.
Bailey, D. W., Gross, J. E., Laca, E. A., Rittenhouse, L. R., Coughenour, M. B., Swift,
D. M., and Sims, P. L. (1996). Mechanisms that result in larger herbivore grazing distribution patterns. Journal of Range Management 49: 386–400.
Bayer, W. (1990). Behavioral compensation for limited grazing time by herded cattle in central
Nigeria. Applied Animal Behaviour Science 27: 9–19.
Birley, M. H. (1982). Resource management in Sukumaland, Tanzania. Africa 52: 1–29.
Borgerhoff Mulder, M. (1998). Brothers and sisters: How sibling interactions affect optimal
parental allocations. Human Nature 9(2): 119–161.
Borgerhoff Mulder, M., and Sellen, D. W. (1994). Pastoralist decision making: A behavioral
ecological perspective. In E. Fratkin, K. A. Galvin, and R. E. A. Roth (eds.), African
Pastoralist Systems: An Integrated Approach. Lynne Reinner Publications, Boulder, CO,
pp. 205–229.
Brandström, P. (1985). The Agro-Pastoral Dilemma: Underutilization or Overexploitation of
Land Among Sukuma of Tanzania. Working Papers in African Studies, African Studies
Programme, University of Uppsala, Uppsala.
Brandström, P., Hultin, J., and Lindström, J. (1979). Aspects of Agropastoralism in East Africa,
Research Report # 51 edition. Scandinavian Institute of African Studies, Uppsala.
Brockington, D., and Homewood, K. (1999). Pastoralism Around Mkomazi: The Interaction
of Conservation and Development. In M. Coe, N. McWilliam, G. Stone, and M. Packer
(eds.), Mkomazi: The Ecology, Biodiversity and Conservation of a Tanzanian Savanna.
Royal Geographic Society (with the Institute of British Geographers), London, pp.
513–530.
Brown, L. H. (1971). The biology of pastoral man as a factor in conservation. Biological
Conservation 3: 93–100.
Burkett, D. W., and Thompson, B. C. (1994). Wildlife association with human-altered water
sources in semiarid vegetation communities. Conservation Biology 8: 682–690.
Landscape Ecology of Herding
557
Charnley, S. (1994). Cattle, Commons and Culture: The Political Ecology of Environmental
Change on a Tanzanian Rangeland. PhD. Dissertation, Stanford University.
Charnley, S. (1997). Environmentally-displaced peoples and the cascade effect: Lessons from
Tanzania. Human Ecology 25: 593–618.
Clutton-Brock, T. H., Guiness, F. E., and Albon, S. D. (1982). Red Deer: The Behaviour and
Ecology of Two Sexes, Chicago University Press, Chicago.
Collett, D. (1989). Pastoralists and wildlife: Image and reality in Kenya Maasailand. In D.
Anderson, and R. Grove (eds.), Conservation in Africa, Cambridge University Press,
Cambridge, pp. 129–167.
Conant, F. P. (1982). Thorns paired, sharply recurved: Cultural controls and rangeland quality
in East Africa. In B. Spooner, and H. S. Mann (eds.), Desertification and Development.
Dryland Ecology in Social Perspective, Academic Press, London, pp. 111–122.
Coppock, D. L., Ellis, J. E., and Swift, D. M. (1986a). Livestock feeding ecology and resource
utilization in a nomadic pastoral ecosystem. Journal of Applied Ecology 23: 573–583.
Coppock, D. L., Swift, D. M., and Ellis, J. E. (1986b). Seasonal nutritional characteristics of
livestock diets in a nomadic pastoral ecosystem. Journal of Applied Ecology 23: 585–595.
Coppolillo, P. B. (in press). Central-place analysis and modeling of landscape-scale resource
use in an East African agropastoral system. Landscape Ecology.
Coughenour, M. B. (1991). Spatial components of plant-herbivore interactions in pastoral,
ranching, and native ungulate ecosystems. Journal of Range Management 44: 530–542.
Dahl, G., and Hjort, A. (1976). Having Herds. Pastoral Herd Growth and Household Economy,
University of Stockholm, Stockholm.
de Boer, W. F., and Prins, H. H. T. (1989). Decisions of cattle herdsmen in Burkina-Faso
and optimal foraging models. Human Ecology 17: 445–464.
du Toit, J. T., and Cumming, D. H. M. (1999). Functional significance of ungulate diversity
in African Savannas and the ecological implications of the spread of pastoralism. Biodiversity and Conservation 8: 1643–1661.
Dyson-Hudson, N., and Dyson-Hudson, R. (1969). Subsistence herding in Uganda. Scientific
American 220: 76–89.
Dyson-Hudson, R., and Dyson-Hudson, N. (1980). Nomadic pastoralism. Annual Reviews in
Anthropology 9: 15–61.
Ellis, J. E., and Swift, D. M. (1988). Stability of African pastoral ecosystems: Alternate
paradigms and implications for development. Journal of Range Management 41: 450–459.
Enghoff, M. (1990). Wildlife conservation, ecological strategies and pastoral communities.
Nomadic Peoples 25–27: 93–107.
Estes, R. D. (1991). The Behavior Guide To African Mammals, University of California
Press, Berkeley.
Evans-Pritchard, E. E. (1940). The Nuer: A Description of the Modes of Livelihood and
Political Institutions of a Nilotic People. Oxford University Press, New York.
Forman, R. T. T. (1995). Land Mosaics: The Ecology of Landscapes and Regions, Cambridge
University Press, Cambridge, New York.
Fox, J., Yonzon, P., and Podger, N. (1996). Mapping conflicts between biodiversity and human
needs in Lantang National Park, Nepal. Conservation Biology 10: 562–569.
Fratkin, E. (1991). Surviving Drought and Development: Ariaal Pastoralists of Northern Kenya.
Conflict and Social Change Series, Westview Press, Boulder, CO.
Fratkin, E., and Smith, K. (1995). Womens changing economic roles with pastoral sedentarization: Varying strategies in alternate Rendille communities. Human Ecology 23: 433–454.
Galaty, J. G. (1988). Pastoral and agropastoral migration in Tanzania: Factors of economy,
ecology and demography in cultural perspective. In J. W. Bennett and J. R. Bower (eds.),
Production and Autonomy, pp. 163–183.
Grootenhuis, J. G., and Olubayo, R. O. (1993). Disease research in the wildlife-livestock
interface in Kenya. The Veterinary Quarterly 15: 55–59.
Homewood, K. M. (1995). Development, demarcation and ecological outcomes in Maasailand.
Africa 65: 331–350.
Homewood, K. M. (1987). Pastoralism, conservation and the overgrazing controversy. In D.
558
Coppolillo
Anderson and R. Grove (eds.), Conservation in Africa: People Policies and Practice,
Cambridge University Press, Cambridge, pp. 111–128.
Homewood, K. M., and Rodgers, W. A. (1984). Pastoralism and Conservation. Human Ecology
12: 431–441.
Homewood, K. M., and Rodgers, W. A. (1991). Maasailand Ecology: Pastoralist Development
and Wildlife Conservation in Ngorongoro, Tanzania. Cambridge Studies in Applied Ecology and Resource Management, Cambridge University Press, Cambridge.
Isbell, L. A. (1991). Contest and scramble competition: Patterns of female aggression and
ranging behavior among primates. Behavioral Ecology 2: 143–155.
Isbell, L. A., Pruetz, J. D., and Young, T. P. (1998). Movements of Vervets (Cercopithecus
aethiops) and Patas monkeys (Erythrocebus patas) as estimators of food resource size,
density, and distribution. Behavioral Ecology and Sociobiology 42: 123–133.
Johnson, D. H. (1989). Political ecology in the upper Nile—The 20th century expansion of
the pastoral common economy. Journal of African History 30: 463–486.
Kikula, I. S. (1997). Policy Implications on Environment: The Case of Villagisation in Tanzania,
Nordic Africa Institute, Uppsala.
Knight, M. H. (1995). Drought-Related mortality of wildlife in the southern Kalahari and the
role of man. African Journal of Ecology 33: 377–394.
Kremen, C., Merenlender, A. M., and Murphy, D. D. (1994). Ecological monitoring—A vital
need for integrated conservation and development programs in the tropics. Conservation
Biology 8: 388–397.
Lamprey, H. (1983). Pastoralism yesterday and today: The overgrazing problem. In F. Bourliere (ed.), Tropical Savannas, Elsevier, Amsterdam, pp. 643–666.
Lamprey, H. F. (1964). Estimation of the large mammal densities, biomass and energy exchange
in the Taringire game reserve and Masai Steppe in Tanzania. East African Wildlife Journal
1: 1–46.
Lane, C. (1996). Ngorongoro Voices, Forests, Trees and People Programme, Sveriges Lantruksuniversitet.
Lindsay, W. K. (1989). Integrating Parks and Pastoralists: Some Lessons from Amboseli. In
D. Anderson, and R. Grove (eds.), Conservation in Africa. Cambridge University Press,
New York, pp. 149–167.
Little, P. D. (1996). Pastoralism, biodiversity, and the shaping of Savanna landscapes in East
Africa. Africa 66: 37–51.
Mace, R. (1993a). Nomadic pastoralists adopt subsistence strategies that maximize long-term
household survival. Behavioral Ecology and Sociobiology 33: 329–334.
Mace, R. (1993b). Transitions Between Cultivation and Pastoralism in Sub-Saharan Africa.
Current Anthropology 34: 363–382.
Maganga, F. P. (1987). Background to Environmental Problems in Sukumaland: An Annotated
Bibliography, Institute of Resource Assessment, University of Dar es Salaam 70, Tanzania.
Marshall, F. (1990). Origins of specialized pastoral production in East Africa. American
Anthropologist 92: 873–894.
McCabe, J. T. (1985). Livestock management among the Turkana: A social and ecological
analysis of herding in an East African pastoral population. PhD. Dissertation, State
University of New York.
McCabe, J. T. (1990a). Success and Failure: The Breakdown of Traditional Drought Coping
Institutions Among the Pastoral Turkana of Kenya. Journal of Asian and African Studies
25: 146–160.
McCabe, J. T. (1990b). Turkana pastoralism: A case against the tragedy of the commons.
Human Ecology 18: 81–103.
McCabe, J. T. (1994). Mobility and land use, old conceptual problems and new interpretations.
In E. Fratkin, K. A. Galvin, and E. A. Roth (eds.), African Pastoralist Systems: An
Integrated Approach, Lynne Reinner Publications, Boulder, CO., pp. 68–89.
McCabe, J. T., Perkin, S., and Schofield, C. (1992). Can conservation and development be
coupled among pastoral people: An examination of the Maasai of Ngorongoro Crater,
Tanzania. Human Organization 51: 353–366.
Landscape Ecology of Herding
559
McCorkle, C. M. (1992). Plants, Animals and People, Westview special studies in social
political & economic development, Westview Press, Boulder, CO.
McIntyre, S., and Hobbs, R. (1999). A framework for conceptualizing human effects on
landscapes and its relevance to management and research models. Conservation Biology
13: 1282–1292.
Morton, J. (1990). Aspects of Labour in an Agro-Pastoral Economy: The Northern Beja of
Sudan, Overseas Development Institute, London.
Niamir, M. (1990). Community Forestry: Herders’ Decision-Making in Natural Resources
Management in Arid and Semi-Arid Africa. Community Forestry, Food and Agriculture
Organization of the United Nations, Rome.
Nicholson, M. J., and Butterworth, M. H. (1986). A Guide to Condition Scoring of Zebu
Cattle: International Liverstock Centre for Africa, Addis Ababa, Ethiopia.
Nyerges, A. E. (1980). Traditional pastoralism and patterns of rangeland degradation. In
H. N. Le Houérou (ed.), Browse in Africa. The Current State of Knowledge, International
Livestock Center for Africa, Addis Ababa, Ethiopia, pp. 465–471.
Nyerges, A. E. (1982). Pastoralists, flocks and vegetation: Processes of co-adaptation. In B.
Spooner, and H. S. Mann (eds.), Desertification and Development: Dryland Ecology in
Social Perspective, Academic Press, London, New York, pp. 217–247.
Prins, H. H. T. (1992). The pastoral road to extinction: Competition between wildlife and
traditional pastoralism in East Africa. Environmental Conservation 19: 117–123.
Roberts, B. D. (1996). Livestock production, age, and gender among the Keiyo. Human
Ecology 24: 215–230.
Roe, E. M. (1984). Range conditions around water sources in Botswana and Kenya. Rangelands
6: 247–249.
Roth, E. A. (1996). Traditional pastoral strategies in a modern world: An example from
northern Kenya. Human Organization 55: 219–224.
Sandford, S. (1982). Pastoral strategies and desertification: Opportunism and conservatism in
dry lands. In B. Spooner, and H. S. Mann (eds.), Desertification and Development: Dryland
Ecology in Social Perspective, Academic Press, London, New York, pp. 61–80.
Schneider, H. (1957). The Subsistence Role of Cattle Among the Pakot and in East Africa.
American Anthropologist 59: 278–300.
Sellen, D. W. (1999). Polygyny and child growth in a traditional pastoral society—The case
of the Datoga of Tanzania. Human Nature 10: 329–371.
Senft, R. L., Coughenour, M. B., Bailey, D. W., Rittenhouse, L. R., Sala, O. E., and Swift,
D. M. (1987). Large Herbivore Foraging and Ecological Hierarchies. Bioscience 37:
789–799.
Sieff, D. (1997). Herding Strategies of the Datoga Pastoralists of Tanzania: Is Household
Labor a Limiting Factor. Human Ecology 25: 519–544.
Sinclair, A. R. E., and Fryxell, J. M. (1985). The Sahel of Africa: Ecology of a disaster.
Canadian Journal of Zoology 63: 987–994.
Tanzania. (1988). Basic Data: Agriculture and Livestock Sector, 1983/84–1987/88, Planning and
Marketing Division, Ministry of Agriculture and Livestock Development, Dar-es-Salaam.
Tanzania. (1989). Tanzania Sensa, 1988 Population Census: Preliminary Report, Bureau of
Statistics, Ministry of Finance, Economic Affairs and Planning, Dar-es-Salaam.
Tanzania. (1992). Basic Data: Agriculture and Livestock Sector, 1985/86–1990/91, Statistics
Unit, Ministry of Agriculture; Bureau of Statistics, Planning and Marketing Division,
Dar-es-Salaam.
Tanzania. (1994). National Sample Census of Agriculture 1993/94, Tanzania Mainland, Ministry
of Agriculture, Statistics Unit Volume II, Dar-es-Salaam.
Thrash, I. (1998). Impact of large herbivores at artificial watering points compared to that at
natural watering points in Kruger National Park, South Africa. Journal of Arid Environments 38: 315–324.
Turner, M. D. (1998a). Long-term effects of daily grazing orbits on nutrient availability in
Sahelian West Africa: 2. Effects of a phosphorus gradient on spatial patterns of annual
grassland production. Journal of Biogeography 25: 683–694.
Turner, M. D. (1998b). Long-term effects of daily grazing orbits on nutrient availability in
560
Coppolillo
Sahelian West Africa: I. Gradients in the chemical composition of rangeland soils and
vegetation. Journal of Biogeography 25: 669–682.
Turner, M. D. (1999). Labor process and the environment: The effects of labor availability
and compensation on the quality of herding in the Sahel. Human Ecology 27: 267–296.
Urban, D. L., O’Neill, R. V., and Shugart, H. H. (1987). Landscape Ecology. Bioscience
37: 119–127.
Verlinden, A. (1997). Human Settlements and Wildlife Distribution in the Southern Kalahari
of Botswana. Biological Conservation 82: 129–136.
Walker, B. (1993). Rangeland ecology. Understanding and managing change. Ambio 22: 80–87.
Waller, R. (1985). Ecology, migration and expansion in East Africa. African Affairs 86:
347–470.
Ward, D., Ngairorue, B. T., Kathena, J., Samuels, R., and Ofran, Y. (1998). Land degradation
is not a necessary outcome of communal pastoralism in arid Namibia. Journal of Arid
Environments 40: 357–371.
Watcher, T. J., Rawlings, P., and Snow, W. F. (1993). Cattle Migration and Stocking Densities
in Relation to Tsetse-trypanosomiasis challenge in The Gambia. Annals of Tropical
Medicine and Parasitology 87: 517–524.
Western, D. (1975). Water availability and its influence on the structure and dynamics of a
Savannah mammal community. East African Wildlife Journal 13: 265–286.
Western, D. (1994). Ecosystem conservation and rural development: The case of Amboseli.
In D. Western, R. M. Wright, and S. C. Strum (eds.), Natural Connections: Perspectives
in Community-Based Conservation, Island Press, Washington DC., pp. 15–52.
Western, D., and Finch, V. (1986). Cattle and Pastoralism: Survival and Production in Arid
Lands. Human Ecology 14: 77–94.
Westoby, M., Walker, B., and Noy-Meir, I. (1989). Opportunistic management for Rangelands
not at equilibrium. Journal of Rage Management 42: 266–274.